George R. Bousfield

Structural features of mammalian gonadotropins

There are two species for which both pituitary and placental gonadotropins are readily available, humans and horses. The human gonadotropins are better characterized than equine gonadotropins. Nevertheless, the latter are very interesting because they provide exceptions to some of the general structure-function principles derived from studies on human and other mammalian gonadotropins. For example, separate genes encode the hLH beta and hCG beta subunits while a single gene encodes eLH beta and eCG beta. Thus, eCG and eLH differ only in their oligosaccharide moieties and eLH is the only LH that possesses the O-glycosylated C-terminal extension previously believed to be restricted to chorionic gonadotropins. Truncation experiments involving eLH beta and hCG beta have suggested the C-terminal extension has no effect on receptor binding. However, the largest of three eCG forms which differ only in the extent of O-glycosylation possessed reduced affinity for LH and FSH receptors. This result suggested that effects of O-glycosylation need to be considered when examining the glycosylation differences between eLH and eCG responsible for the 10-fold lower eCG receptor binding affinity compared with that of eLH. Contribution of alpha Asn56 N-linked oligosaccharides to the different biological activities of eLH and eCG has been evaluated following selective removal using peptide-N-glycanase digestion of native equine alpha-subunit preparations. Hormones-specific patterns of glycosylation were observed on alpha Asn56 of eLH, eFSH, and eCG. Removal of alpha Asn56 oligosaccharides increased the rate of subunit association, the extent of association, and receptor binding activity. Some unassociated alpha-subunit oligosaccharides were identified which may interfere with subunit association because they were more abundant in unassociated subunit oligosaccharide maps than in a total oligosaccharide map. This was most striking in the case of eCG alpha in which two minor peaks became the major oligosaccharide peaks detectable in the unassociated eCG alpha fraction following association with eLH beta and eFSH beta. The biological activities exhibited by hybrid hormones, eLH alpha reassociated with oLH beta and pLH beta, found to be greater than those of oLH and pLH provided an interesting exception to the general rule that the beta-subunit determines the potency of the heterodimer. LH receptor binding activities of eLH beta-chimeric ovine/equine alpha-subunits suggested that the equine alpha-subunit N-terminal domain may be responsible for this effect. Equine FSH has higher FSH receptor binding activity than human, ovine, and porcine FSH preparations. This probably results from two factors. First, the presence of the equine alpha-subunit promotes receptor binding as noted above. Second, the overall -2 charge of the eFSH beta determinant loop, which is less negative that the -3 observed in other species, results from the presence of an Asn residue at position 88 instead of Asp. This apparently facilitates binding to the FSH receptor.

All-or-none N-glycosylation in primate follicle-stimulating hormone β-subunits

Abstract Human FSH exists as two major glycoforms designated, tetra-glycosylated and di-glycosylated hFSH. The former possesses both α- and β-subunit carbohydrates while the latter possesses only α-subunit carbohydrate. Western blotting differentiated the glycosylated, 24,000 M r hFSHβ band from the non-glycosylated 21,000 M r FSHβ band. Postmenopausal urinary hFSH preparations possessed 75–95% 24,000 M r hFSHβ, while pituitary hFSH immunopurified from 21- to 43-year-old females and 21–43-year-old males possessed only 35–40% 24,000 M r hFSHβ. The pituitary hFSH from a postmenopausal woman on estrogen replacement was 75% 21,000 M r hFSHβ. Other immunopurified postmenopausal pituitary hFSH preparations possessed 50–60% 21,000 M r hFSHβ. Gel filtration removed predominantly 21,000 M r free hFSHβ and reduced its abundance to 13–22% in postmenopausal pituitary hFSH heterodimer preparations. A major regulatory mechanism for FSH glycosylation involves control of β-subunit N -glycosylation, possibly by inhibition of oligosaccharyl transferase. Two primate species exhibited the same all-or-none pattern of pituitary FSHβ glycosylation.

Partially Deglycosylated Equine LH Preferentially Activates β-Arrestin-Dependent Signaling at the Follicle-Stimulating Hormone Receptor

Deglycosylated FSH is known to trigger poor Galphas coupling while efficiently binding its receptor. In the present study, we tested the possibility that a deglycosylated equine LH (eLHdg) might be able to selectively activate beta-arrestin-dependent signaling. We compared native eLH to an eLH derivative [i.e. truncated eLHbeta (Delta121-149) combined with asparagine56-deglycosylated eLHalpha (eLHdg)] previously reported as an antagonist of cAMP accumulation at the FSH receptor (FSH-R). We confirmed that, when used in conjunction with FSH, eLHdg acted as an antagonist for cAMP accumulation in HEK-293 cells stably expressing the FSH-R. Furthermore, when used alone at concentrations up to 1 nM, eLHdg had no detectable agonistic activity on cAMP accumulation, protein kinase A activity or cAMP-responsive element-dependent transcriptional activity. At higher concentrations, however, a weak agonistic action was observed with eLHdg, whereas eLH led to robust responses whatever the concentration. Both eLH and eLHdg triggered receptor internalization and led to beta-arrestin recruitment. Both eLH and eLHdg triggered ERK and ribosomal protein (rp) S6 phosphorylation at 1 nM. The depletion of endogenous beta-arrestins had only a partial effect on eLH-induced ERK and rpS6 phosphorylation. In contrast, ERK and rpS6 phosphorylation was completely abolished at all time points in beta-arrestin-depleted cells. Together, these results show that eLHdg has the ability to preferentially activate beta-arrestin-dependent signaling at the FSH-R. This finding provides a new conceptual and experimental framework to revisit the physiological meaning of gonadotropin structural heterogeneity. Importantly, it also opens a field of possibilities for the development of selective modulators of gonadotropin receptors.

Hypo-glycosylated human follicle-stimulating hormone (hFSH21/18) is much more active in vitro than fully-glycosylated hFSH (hFSH24)

Hypo-glycosylated hFSH(21/18) (possesses FSHβ(21) and FSHβ(18)bands) was isolated from hLH preparations by immunoaffinity chromatography followed by gel filtration. Fully-glycosylated hFSH(24) was prepared by combining the fully-glycosylated FSHβ(24) variant with hCGα and isolating the heterodimer. The hFSH(21/18) glycoform preparation was significantly smaller than the hFSH(24) preparation and possessed 60% oligomannose glycans, which is unusual for hFSH. Hypo-glycosylated hFSH(21/18) was 9- to 26-fold more active than fully-glycosylated hFSH(24) in FSH radioligand assays. Significantly greater binding of (125)I-hFSH(21/18) tracer than hFSH(24) tracer was observed in all competitive binding assays. In addition, higher binding of hFSH(21/18) was noted in association and saturation binding assays, in which twice as much hFSH(21/18) was bound as hFSH(24). This suggests that more ligand binding sites are available to hFSH(21/18) in FSHR than to hFSH(24). Hypo-glycosylated hFSH(21/18) also bound rat FSHRs more rapidly, exhibiting almost no lag in binding, whereas hFSH(24) specific binding proceeded very slowly for almost the first hour of incubation.

Identification of Twelve O-Glycosylation Sites in Equine Chorionic Gonadotropin β and Equine Luteinizing Hormone β by Solid-Phase Edman Degradation

Abstract The O-glycosylation sites for equine LHβ (eLHβ) and eCGβ were identified by solid-phase Edman degradation of four glycopeptides derived from the C-terminal region. Both subunits were O-glycosylated at the same 12 positions, rather than the 4–6 sites anticipated. These sites were partially glycosylated, with carbohydrate attachment ranging from 20% to 100% for eCGβ and from 10% to 100% for eLHβ. When the C-terminal peptide containing all but one of the O-linked oligosaccharides was removed by mild acid hydrolysis of either eLHβ or eCGβ, hybrid hormones could be obtained by reassociating eLHα,eFSHα, or eCGα with the truncated β subunit derivatives. These hybrid hormones were identical in LH receptor-binding activity when des(121-149)eLHβ or des(121-149)eCGβ were combined with the same α subunit preparation. Thus, O-glycosylation appears to be responsible for the β subunit contribution to the substantial difference in LH receptor-binding activity between eLH and eCG. Comparison of the equid LH/CGβ sequences with those available for the primate CGβ subunits indicated a greater conservation of glycosylation patterns in the former.

Synthesis and secretion of gonadotropins including structure-function correlates

The synthesis and secretion of the gonadotropic hormones involves coordination of signal transduction, gene expression, protein translation, post-translational folding and modification and finally secretion. The production of biologically active gonadotropin thus requires appropriately folded and glycosylated subunits that assemble to form the heterodimeric hormone. Here we overview recent literature on regulation of gonadotropin subunit gene expression and current understanding of the assembly and secretion of biologically active gonadotropic hormones. Finally, we discuss the therapeutic potential of understanding glycosylation function towards designing new forms of gonadotropins based on observations of physiologically relevant parameters such as age related glycosylation changes.

Chromatofocusing fails to separate hFSH isoforms on the basis of glycan structure.

Follicle-stimulating hormone (FSH) glycosylation is regulated by feedback from the gonads, resulting in an array of glycans associated with FSH preparations derived from pools of pituitary or urine extracts. FSH glycosylation varies due to inhibition of FSHbeta N-glycosylation, elaboration of 1-4 branches possessed by mature N-glycans, and the number and linkage of terminal sialic acid residues. To characterize FSH glycosylation, FSH isoforms in pituitary gland extracts and a variety of physiological fluids are commonly separated by chromatofocusing. Variations in the ratios of immunological and biological activities in the resulting FSH isoform preparations are generally attributed to changes in glycosylation, which are most often defined in terms of sialic acid content. Using Western blotting to assess human FSHbeta glycosylation inhibition revealed 30-47% nonglycosylated hFSHbeta associated with four of six hFSH isoform preparations derived by chromatofocusing. Glycopeptide mass spectrometry assessment of glycan branching in these isoforms extensively characterized two N-glycosylation sites, one at alphaAsn52, the critical glycan for FSH function, and the other at betaAsn24. With two to four N-glycans per FSH molecule, many combinations of charges distributed over these sites can provide the same isoelectric point. Indeed, several glycans were common to all isoform fractions that were analyzed. There was no trend showing predominantly monoantennary glycans associated with the high-pI fractions, nor were predominantly tri- and tetra-antennary glycans associated with low-pI fractions. Thus, differences in receptor binding activity could not be associated with any specific glycan type or location in the hormone. FSH aggregation was associated with reduced receptor binding activity but did not affect immunological activity. However, as gel filtration indicated sufficient heterodimer was present in each isoform preparation to generate complete inhibition curves, the near total loss of receptor binding activity in several preparations could not be explained by aggregation alone, and the mechanism remains unknown.

Production, purification, and characterization of recombinant hFSH glycoforms for functional studies

Previously, our laboratory demonstrated the existence of a β-subunit glycosylation-deficient human FSH glycoform, hFSH(21). A third variant, hFSH(18), has recently been detected in FSH glycoforms isolated from purified pituitary hLH preparations. Human FSH(21) abundance in individual female pituitaries progressively decreased with increasing age. Hypo-glycosylated glycoform preparations are significantly more active than fully-glycosylated hFSH preparations. The purpose of this study was to produce, purify and chemically characterize both glycoform variants expressed by a mammalian cell line. Recombinant hFSH was expressed in a stable GH3 cell line and isolated from serum-free cell culture medium by sequential, hydrophobic and immunoaffinity chromatography. FSH glycoform fractions were separated by Superdex 75 gel-filtration. Western blot analysis revealed the presence of both hFSH(18) and hFSH(21) glycoforms in the low molecular weight fraction, however, their electrophoretic mobilities differed from those associated with the corresponding pituitary hFSH variants. Edman degradation of FSH(21/18)-derived β-subunit before and after peptide-N-glycanase F digestion confirmed that it possessed a mixture of both mono-glycosylated FSHβ subunits, as both Asn(7) and Asn(24) were partially glycosylated. FSH receptor-binding assays confirmed our previous observations that hFSH(21/18) exhibits greater receptor-binding affinity and occupies more FSH binding sites when compared to fully-glycosylated hFSH(24). Thus, the age-related reduction in hypo-glycosylated hFSH significantly reduces circulating levels of FSH biological activity that may further compromise reproductive function. Taken together, the ability to express and isolate recombinant hFSH glycoforms opens the way to study functional differences between them both in vivo and in vitro.

Hypoglycosylated hFSH Has Greater Bioactivity Than Fully Glycosylated Recombinant hFSH in Human Granulosa Cells

CONTEXT Previous studies suggest that aging in women is associated with a reduction in hypoglycosylated forms of FSH. OBJECTIVE Experiments were performed to determine whether glycosylation of the FSHβ subunit modulates the biological activity of FSH in human granulosa cells. DESIGN AND SETTING Recombinant human FSH (hFSH) derived from GH3 pituitary cells was purified into fractions containing hypoglycosylated hFSH(21/18) and fully glycosylated hFSH(24). The response to FSH glycoforms was evaluated using the well-characterized, FSH-responsive human granulosa cell line, KGN at an academic medical center. INTERVENTIONS Granulosa cells were treated with increasing concentrations of fully- or hypoglycosylated FSH glycoforms for periods up to 48 hours. MAIN OUTCOME MEASURE(S) The main outcomes were indices of cAMP-dependent cell signaling and estrogen and progesterone synthesis. RESULTS We observed that hypoglycosylated FSH(21/18) was significantly more effective than fully glycosylated FSH(24) at stimulating cAMP accumulation, protein kinase A (PKA) activity, and cAMP response element binding protein (CREB) (S133) phosphorylation. FSH(21/18) was also much more effective than hFSH(24) on the stimulation CREB-response element-mediated transcription, expression of aromatase and STAR proteins, and synthesis of estrogen and progesterone. Adenoviral-mediated expression of the endogenous inhibitor of PKA, inhibited FSH(21/18)- and FSH(24)-stimulated CREB phosphorylation, and steroidogenesis. CONCLUSIONS Hypoglycosylated FSH(21/18) has greater bioactivity than fully glycosylated hFSH(24), suggesting that age-dependent decreases in hypoglycosylated hFSH contribute to reduced ovarian responsiveness. Hypoglycosylated FSH may be useful in follicle stimulation protocols for older patients using assisted reproduction technologies.

Comparison of Follicle-Stimulating Hormone Glycosylation Microheterogenity by Quantitative Negative Mode Nano-Electrospray Mass Spectrometry of Peptide-N-Glycanase-Released Oligosaccharides

Glycans from six highly purified hFSH preparations were released by peptide-N-glycanase digestion and analyzed by negative mode nano-ESI mass spectrometry before and after neuraminidase digestion. Pituitary glycan structures were mainly high-mannose, di-, tri-, and tetra-antennary, and their abundance largely paralleled that reported by other investigators using different approaches. For most of the FSH preparations, the differences in glycosylation appeared to be restricted to relative abundances of the major glycan families, as defined by their neutral core oligosaccharide structures. Qualitative differences between glycan populations were largely relegated to those species that were lowest in abundance. Significant qualitative differences were noted in two cases. Recombinant GH3-hFSH triantennary glycans appeared to have the third antenna exclusively on the mannose6-branch, in contrast to all pituitary and urinary hFSH triantennary glycans, in which this antenna was exclusively attached to the mannose3-branch. The hypo-glycosylated hFSH preparation isolated from purified hLH was decorated with high mannose glycans that accounted for over 40% of the total in this population. As this preparation was found to be consistently 20-fold more active than hFSH24 in FSH receptor-binding assays, it appears that both macroheterogeneity and microheterogeneity in FSH preparations need to be taken into account.